UNCLASSIFIED

Defense Technical Information CenterCompilation Part Notice

ADPO11198TITLE: Nanoscale Variation in Electric Potential at Oxide Bicrystal andPolycrystal Interfaces

DISTRIBUTION: Approved for public release, distribution unlimited

This paper is part of the following report:

TITLE: Internal Workshop on Interfacially Controlled FunctionalMaterials: Electrical and Chemical Properties Held in Schloss Ringberg,Germany on March 8-13, 1998

To order the complete compilation report, use: ADA397655

The component part is provided here to allow users access to individually authored sectionsf proceedings, annals, symposia, etc. However, the component should be considered within

[he context of the overall compilation report and not as a stand-alone technical report.

The following component part numbers comprise the compilation report:ADPO11194 thru ADP011211

UNCLASSIFIED

NHSOLID

STATEELSEVIER Solid State Ionics 131 (2000) 51-60 IONICS

www.elsevier.com/locate /ssi

Nanoscale variation in electric potential at oxide bicrystal andpolycrystal interfaces

Bryan D. Huey, Dawn A. Bonnell*The University of Pennsylvania, Department of Materials Science, Philadelphia, PA 19104, USA

Received 1 September 1999; accepted 1 December 1999

Abstract

Scanning surface potential microscopy (SSPM), has been used to measure spatial variations in grain boundary propertiesin SrTiO3 and ZnO. Experimental measurements of a Fe-doped SrTiO3 13 bicrystal are compared to finite elementcalculations to quantify the effects of tip geometry and sample-tip separation. Experimental and numerical treatments includerealistic tip interactions and lateral inhomogeneity in sample properties. A procedure for extracting actual interface potentialsfrom separation dependence is proposed. Both the sign and magnitude of the grain boundary potential barrier measured withSSPM agree with macroscopic measurements. For experimentally available tips, the effect of tip geometry was found not tocontribute to uncertainty. In application to polycrystalline materials, the voltage dependence of individual interface propertieshas been determined in micropattemed, ZnO-based, polycrystalline varistor devices. © 2000 Elsevier Science BV. Allrights reserved.

Keywords: Atomic force microscopy; Surface potential; SSPM; Interface potential; Oxide

1. Introduction achieved using either transmission electron micro-scopy (TEM) or probe-based measurements. For

To support continued miniaturization and hybridi- example, scanning tunneling microscopy (STM) haszation of oxide electronic devices, a fundamental been used to detect variations in conductivity atunderstanding of interfacial effects at the nanometer polycrystalline ZnO grain boundaries [4], and toscale is necessary. It is at this spatial scale that measure Schottky barrier formation at the interfacespotential barriers exist due to charge trapped at between TiO 2 and metal clusters [5]. Electrostaticinterfaces, facilitated by segregation of dopants and force detection based on atomic force microscopyimpurities [1] as well as the presence of thin (AFM) at in situ biased Nb-doped TiO 2 polycrystalsamorphous layers [2,3]. Such scales are experimen- observed the voltage dependence of grain boundarytally difficult to access, and relevant property mea- potential barriers [6]. Potential variations were quan-surements at oxide interfaces have only been tified using scanning surface potential microscopy

(SSPM), in an investigation of ZnO varistor materi-

*Corresponding author. Tel.: + 1-215-898-6231; fax: + 1-215- als also with in situ applied biases [7]. This tech-573-2128. nique was applied by Nabhan et al. [8] to directly

E-mail address: bonell@soll.lrsm.upenn.edu (D.A. Bonnell) observe potential barriers in acceptor p-type poly-

0167-2738/00/$ - see front matter © 2000 Elsevier Science BV. All rights reserved.PII: S0167-2738(00)00621-4

52 B.D. Huev, D.A. Bonnell / Solid State lonics 131 (2000) 51-60

crystalline silicon. Vandervorst et al. [9] applied a carrier concentration is 2.6 X 1017 charges/cm3)scanning probe tip as a contact voltage probe to [15]. The ideal atomic structure of the grain bound-detect potential variations in CMOS transistors. For ary core was measured with high resolution transmis-the particular case of SrTiO3 bicrystal interfaces sion electron microscopy for similar samples (1.3 XMaier et al. [10,11] have investigated tilt and twist 1019 Fe atoms/cm3 23 {111} bicrystals); the inter-acceptor-doped SrTiO. bicrystal interfaces at tem- face was found to consist of Sr and 0 atoms at theperatures above 500'C using macroscopic impedance boundary plane with perfect mirror symmetry [16].measurements and in situ electrocoloration. Dravid et The bicrystal used herein was polished with diamondal. [12] employed holographic phase contrast in impregnated films down to 0.05 pim grit size. ToTEM to estimate the charge and potential barrier apply local lateral electrostatic fields to the interface,associated with bicrystal grain boundaries, and to a grid of 400 nm thick, 100 pLm diameter Al contactsobserve the disappearance of the barrier with the separated by 350 pxm gaps was electron beamapplication of large external biases [13]. However, evaporated through a shadow mask onto the surfacedifficulties in interpretation and sample preparation of the bicrystal.for in situ TEM studies suggest that probe-based Individual grain boundaries in commercial ZnO-techniques may be better suited for the investigation based polycrystalline varistors [17] were isolatedof samples of interest to the device fabrication using microlithography. Contacts were deposited oncommunity. these samples using standard lift-off metallization

Since the scanning techniques of surface potential described in detail elsewhere [7]. The measurementmeasurement are new and in situ measurements have is considered in situ because micropatterning allowsonly been accomplished recently, careful considera- external lateral biases to be applied at individualtion of tip-sample interactions and resolution limits grain boundaries during SSPM. Biases were appliedbased on model experiments is required. This paper with a function generator (Wavetek model 20), andfirst compares surface potential measurements of current was monitored with a multimeter (Keithlybicrystal interfaces by SSPM with numerical calcula- model 175) connected in series with the sample.tions. The spatial variation of the interface potential The configuration of local electrostatic potentialbarrier of well characterized SrTiO 3 bicrystals pro- variation near an interface is described as follows. Avides the model system. Finally, individual grain two dimensional defect such as a grain boundary canboundaries of polycrystalline ZnO-based commercial trap charge, and in an oxide the consequence to localvaristors were investigated under in situ applied field variation extends over a large range due to thefields, allowing determination of the voltage depen- magnitude of the dielectric constants. Using tradi-dence of grain boundary electronic properties and tional semiconductor models, a grain boundary inter-demonstrating the general application of the ap- face charge can be treated as a delta function, withproach. an oppositely charged depletion region extending

into the adjacent grains so that charge is conserved.Fig. l(a), left, depicts a negative grain boundary

2. Experimental details charge (infinitesimally thin) as well as a positivedepletion region (in extinction and using the abrupt

A bicrystal presents an ideal geometry with which junction approximation). Solving Poisson's equationto investigate interactions between a tip and a true for this charge density distribution yields the equilib-sample surface. A 70.6' pure tilt 13 grain boundary rium energy band diagram indicated in Fig. 1 (b), left,along the {111} plane in acceptor-doped SrTiO 3 for a n-type semiconductor. In the event that a(4.8 X 10's Fe atoms/cm3 ) was fabricated using the positive bias (V) is applied to the right grain whileVerneuil method [14]. Macroscopic conductivity and the left grain is grounded, the depletion regioncapacitance measurements at elevated temperatures lengthens or shortens for the 'reverse' and 'forward'(500-1000°C) yielded average values for the grain biased grains, respectively. The interface charge canboundary potential barrier and depletion width of also increase, if empty interface states above the300 mV and 400 nm, respectively (the activated Fermi level are available. The charge distribution for

B.D. Huey, D.A. Bonnell / Solid State lonics 131 (2000) 51-60 53

Sample Tip Sample (+V at right)

a w-- . + I w+

charge charge,density ,density

b eV , eV(.. . WFt-WFs)-----------WFs Evac - -E-vacvCLtEvac \-Eva *

+ * gb, Ec IWFt, tgbS...... ........................................E Ef.:•;•~g;)t-.•E't5.>••;Z' 42'";/~~•/: ............. it ý• Ef

A Ef

Ev

CjeV SSpM e L/ýSý

Fig. 1. The relationship between (a) charge density, (b) energy band diagram, and (c) measured surface potential for a grounded npnjunction or polycrystalline grain boundary in thermal equilibrium with the SSPM tip (center). The left and right sides, respectively, depictthe model without and with an external bias (V) applied to the right side. Parameters include: w = depletion width; 8 = thickness of p region(---40); Ogb = potential barrier height; WF, and WF, = work function of sample and tip, respectively; and Eva.. E•, Ef, and E = vacuum level,conduction band, Fermi level, and valence bands.

this biased case is shown at the far right of Fig. l(a), where the work functions of tip and sample (WF) arewhile the right of Fig. l(b) indicates the modified considered:band diagram for an external applied bias. The VV sample

potential barrier broadens, becomes asymmetric, and iWdecreases in magnitude. = (VNull - Vsample - (WFTip - WFsampie))

To measure the voltage dependence of potential + VAC sin(wt) = VDC + V,, sin(wt)barriers, SSPM was employed. SSPM is based on thecentury old concept of a Kelvin probe [18] or nullforc prbe [9-2]. ensiiviy t thesurace The capacitive force exerted on the tip above thefo rc e p ro b e [19 - 2 2 ]. S e n sitiv ity to th e su rfac es u f c i s t e p r o t o n l o th sq a e f A Vpotential of a sample is achieved by nulling the surface is the pootinal to t se of ( V,capacitive force acting on a biased Permalloy coated resulting in three distinct force components (DC, w,AFM tip oscillating above a surface. Using standard a n an) whr , is the separationcbetweenptiptandintermittent contact AFM, the grounded tip first specimen and t is the total effective capacitanceacquires a trace of the surface topography in the uing teip apex, tip edges, and cantilevervicinity of a grain boundary intersecting a surface.The tip then retraces the topographic profile, sepa- 1 aCEff

rated from the surface 50-100 nm, thereby maintain- F 2 azing a constant tip-sample separation. During the 2 1 2 12

second scan, a variable DC bias (VNI,) is applied to [ VDC + 2 VAC) - 2 VAC cos(2ot)the tip (Vp) as well as a fixed frequency AC bias(VAc" sin(&ot)). This leads to a potential difference + 2VDcVAc sin(wt)|J(AV) between tip and an arbitrarily biased sample(Vsample) that includes DC and AC components, The force component acting on the tip (and causing

54 B.D. Huey, D.A. Bonnell / Solid State Ionics 131 (2000) 51-60

it to oscillate) at (o will be eliminated (null) when the potential, and work function for the surface regionadjustable tip bias and the sample surface potential 'i,' respectively, and VNuU.H,t is the measured surfaceare equal (VDC =0), offset by the differences in work potential by the SSPM tip) [23]function:

VNuI = Vs,,,l... + (WFT,p - WFs1,1pie) F1 ,,, - az VDC C = VAC , LI VI c

In this manner, the surface potential at any position "1 aCof the surface is determined by recording the adjust- = Vtc. 1 [VN1,,I.H,, - VYs - (WFripable tip bias (VN,,,I) that nulls the tip oscillation at to. iF)

Fig. 1(c) depicts the anticipated SSPM profiles for -WFs,)]

the grounded and biased cases of the grain boundary Solving for F=0, the analytical solution for thewith negative interface charge considered in Fig. 1(a) SSPM measured surface potential of a heterogeneousand (b). As can be seen, SSPM yields the surface sample (VKI.H.) is:potential, which is the mirror image of the energyband diagram, offset by the tip-sample work func- aCEfr•tion difference. It is important to note that the tip and Z [s + (WF1,5 - WF5 )]

sample apparent work functions can be affected by VN,,l.Ht = a 8Crrsurface adsorption. However, assuming that thiseffect is constant with lateral position, the measured

surface potential is offset by a constant and spatial These relations are used in the comparison ofvariations are valid. experimental results and numerical calculations.

This simplified description of SSPM does not takeinto account capacitive interactions between thebiased tip and a heterogeneous surface potential in 3. Results and discussionthe sample. Fig. 2 depicts this case, including sixdistinct regions of differing surface potential. Fol- The topographic structure of the 1:3 SrTiO3 grainlowing Henning et al. [5], the equation that describes boundary is shown in Fig. 3(a), where only polishingthe force acting on the tip at wo for n distinct surface damage is visible. Fig. 3(b) reveals a variation in the

regions is thus revised (where CEtfi, Vsi, and WFs.1 surface potential measured with a tip height of 80 nm

represent the ith components of capacitance, surface that takes the form of a ridge of raised surfacepotential. The contrast variation is more visible inFig. 3(c). The average height of this feature is 25 mV

(95% confidence is ±9 mV), and the depletion widthis 2770±40 nm. It is noteworthy that the grainboundary is not visible in the topography; theposition was confirmed ex situ by electron backscat-tering (EBSP) in a scanning electron microscope(SEM) [24]. Note that the feature clearly varies withposition, ranging from 11 to 39 mV in height and2.150-3.520 pRm in width (in excess of the ex-perimental energy noise of 5 mV). Furthermore, theapparent' interface potential and depletion width

depend on the height at which the image wasobtained. Table I summarizes 'apparent' properties

Fig. 2. Schematic diagram of how a heterogeneous potential at several heights.distribution on a sample surface contributes to the capacitancemeasured by a tip. Regions directly below the tip affect a smaller The presence of increased surface potential at thetip area than do surrounding regions, resulting in a non-linear grain boundary position can only result from a

interaction. depression in the corresponding band diagram, the

B.D. Huey, D.A. Bonnell / Solid State lonics 131 (2000) 51-60 55

-40.0 -40.0

-30.0 -30.0

b

020.0 0 0 .20.0 302 U

T10.0 1i0.0

0 10.0 nm i0Dl 10o 0 20.0 30.0 40.0 0 10.0 20.0 30.0 40.0

C

Fig 3.Toogrphi srucure(a an sufae ptenia inpl n veou potad3- entildisributic)on at SThe grain boundary.

iepnverse of Fig.en 1(),asepe ctiedo for angh acetr- arfoprtie to0] qu sdiceantiyteeefcts two-bedimensiona

dopedrmacepteriln Aptnti alV4+ troug indcaiv of efiniteieemnofth calculations [25 dsrthel prsolvngeoDpositive winterface charge170impedes0current 30flowbe-haPoisson'steq uation foote x erientotaltip/sample

cause positive charge carriers dominate conduction geometry are compared.for the bicrystal in contrast to the n-type varistors. Fig. 4(a) and (b) depict the model used in theThe 'apparent' interface potential is of the same sign simulation, including a tip with a 50 am radius ofbut is approximately one order of magnitude lower curvature, a tip length of 12.5 jim, a tip half angle of

56 B.D. Huey, D.A. Bonnell / Solid State lonics 131 (2000) 51-60

a C

/%

________ __\_ NýT

Fig. 4. Two dimensional finite element calculations of a charged grain boundary intersecting a surface. The model includes tip(length= 12.5 lim, half angle= 17.50) and cantilever width=25 pRm. (a) The details in the tip region (b) show the interface which has acharge of + I X 1013 e/cm 2, and adjacent depletion regions with -2.6X 1017 e/cm3 . Solutions for the electric potential with the tip over theinterface (c) and displaced laterally (d) are compared.

17.5', a cantilever width of 25 pm, a tip-sample lifted directly above the grain boundary, in which theseparation of 60 nm, and a grain boundary with the contrast represents the potential distribution rangingmacroscopic potential barrier oriented perpendicular between 0 and 300 mV (darkest to lightest). Ato the surface. Boundary conditions include a posi- parabolic potential barrier exists at the interface astive interface charge of + 1.0X 1013 charges/cm2 for anticipated. The tip experiences a net null force inthe centered grain boundary, a potential barrier of the simulation following the procedure of Jacobs et300 mV, and symmetric, negatively charged depletion al. [26]. By repeating calculations with the tip atregions extending 200 nm into each of the adjacent successively larger lateral displacements from thegrains with an ionized acceptor concentration of grain boundary, as exemplified in Fig. 4(d), a-2.6 X 10'" charges/cm3 . The number of nodes at potential barrier profile simulating the SSPM experi-which Poisson's equation is discretely solved is 485, ment is obtained. Fig. 5 depicts the calculations forwhere at least 25 nodes are always present between the experimental tip geometry, as well as a SSPMthe tip apex and the sample (the simulation mesh is measured surface potential profile extracted from aindicated in Fig. 4(b), a close-up of Fig. 4(a)). Fig. potential image acquired 60 nm above the surface.4(c) shows the simulation in the vicinity of the tip Note that the calculated potential maximum agrees

B.D. Huey, D.A. Bonnell Solid State lonics 131 (2000) 51-60 57

this effect, numerical calculations with models simi-lar to that in Fig. 4 were done for the most 'ideal'

""A0 shapes available as conducting tips [27]. TheseE A calculations showed a negligible dependence of the

potential barrier height on tip geometry at heights in

0) X, the range of 60-100 rnm; differences of 10-20% inA apparent feature width might be determined for tips

. 20 A with radii of curvature between 25 nm up to an upperz A A 10 A•. Alimit of 200 nm. A procedure for extracting theZ 10 actual value from the measurements has not yet been

- O- 4 developed.

-500 0 500 The remaining differences between propertiesdetermined from macroscopic transport studies and

Lateral Distance to GB (nm) local probe measurements may be related to sample

Fig. 5. Comparison of experimental (cross hatches) and calculated conditions. It is known that oxide surfaces adsorb(triangles) potential profiles at 60 nm above the surface. The atmospheric molecules that alter the surface chargeaverage experimentally measured potential maximum was 46±5 [28]. The presence of surface charge might dampenmV. or heighten lateral potential variations. Preliminary

simulations accounting for surface charge and associ-ated surface band bending suggest this effect is

with the average measured value, 46 mV, although negligible. It is also possible that the interfacethe particular profile compared in the figure is in the intersects the surface at an angle that deviates fromupper range of the experimental distribution. 900. Relative misorientation determined by EBS is

Of course electric fields decay with distance from less than ±50 and would not account for significantthe surface so the measured valued depends on the variations in depletion width. Finally, the bicrystalheight at which it was obtained. To extract the actual grain boundary may not be as perfect macroscopical-value at the surface, the distance dependence of ly as suggested by TEM. Amorphous interfacialresults from the numerical calculations is parame- films, facets, and pores are likely to exist along theterized, thus accounting for the inhomogeneous boundary. This possibility is supported by recentsurface potential and realistic tip dimensions. The near field optical studies on similar bicrystals andideal relation, i.e. that from the calculations, can be thin films [29].described as: With this insight, polycrystalline materials can be

quantified from several perspectives. Fig. 6(a) andkz = P, exp[Bzc] (b) compare the topography and surface potential of

the ZnO varistor when leads just to the left and rightwhere 0, is the potential measured at some height z, of the imaged region are grounded. Again, a grainq0 is the potential at the interface, B and C are boundary is not apparent in the topographic struc-constants that characterize the complex sample-tip ture. Two regions of depressed surface potential existinteraction, equal -0.47±0.01 and 0.345±0.005, in Fig. 6(b) (-70 mV), resulting from the presencerespectively. Using this functional form to determine of second phases with a larger apparent workthe interface potential from the measurements yields function than that of the surrounding grains (includ-260±10 mV, in good agreement with macroscopic ing the mediation of the sample work function bymeasurements, surface adsorbates). These regions have been con-

The 'apparent' depletion depth measured with firmed to be Bi and Ti rich second phases by energySSPM is larger than expected from macroscopic dispersive spectroscopy in SEM [7]. Fig. 6(c) and (d)transport measurements. Tip geometry is an obvious depict the surface potential distribution when a biasand well known factor in spatial resolution degra- of ± 1 V is externally applied to the left contact,dation in scanning probe microscopies. To quantify respectively, causing current flow through the var-

58 B.D. Huey, D.A. Bonnell / Solid State Ionics 131 (2000) 51-60

...500.0 n ,

?5F . 0

C) 0 rs•,

Fig. 6. Topographic structure (a) and surface potential of unbiaed region of a polycrystaIlne ZnO surfacc. Surface potentia variton at a

grain boundary with laterally applied (c) forward and (d) reverse bias (±1.00 V applied approximately 30 urn to the left and right).

istor during this in situ SSPM experiment. The sharp, mV exists along the entire grain boundary correlating

irregular contrast step in the center of these surface to the difference in relative effective work functions

potential images represents a highly resistive feature, of the adjacent grains.resulting from the presence of a potential barrier Extracting surface potential profiles at a given

containing grain boundary. The height of the po- position for a series of external biases, and knowing

tential step is approximately 190 mV, although it the current flowing through the device during the invaries as a function of position (220-160 mV from situ SSPM measurements, the voltage dependence oftop to bottom of the 10 p2m grain boundary). The the potential barrier can be determined. Current is

potential step also inverts when the polarity of the assumed to flow by thermionic emission over the

external bias is reversed, as expected, except at the potential barrier (EF refers to the Fermi energy with

2nid phases where the electrically active interface respect to the conduction band minimum; k is

switches from one side to the other of the impurity Boltzman's constant; T is room temperature; l/Arearegion. Reanalyzing Fig. 6(b), an average step of 7 is the measured current density flowing through the

B.D. Huey, D.A. Bonnell / Solid State lonics 131 (2000) 51-60 59

leads during the in situ experiment (the cross sec- boundary differ substantially from the macroscopic,tional area is taken from the microfabricated contact average response.dimensions); and Vapp is the bias dropping at thegrain boundary measured directly from SSPM pro-files): 4. Conclusion

q5gb(V) = -EF - kT A correlation of scanning probe-based measure-ments of surface potential and actual local properties

In L 2 I/Area has been made for the case of a charged interfaceAT Vaxp intersecting a surface. The potential barrier of an

1 -exp - kT / acceptor-doped, SrTiO3 13 bicrystal grain boundary

is found to correspond to a positively chargedFig. 7 presents the voltage dependence of the interface and a negative depletion charge. There is an

individual potential barrier from Fig. 6 and compares intrinsic experimental problem with measurement atresults from the local and macroscopic measure- the surface so the actual properties must be extractedments. The two curves for the local properties from measurements above the surface. Analyticalrepresent upper and lower bounds for the actual solutions of simple geometries would not provide thepotential barrier directly beneath the tip. Specifically, correct relation between the experiment and prop-the lower bound is obtained by assuming all mea- erties because the tip-sample interactions are com-sured current flows through (over) the potential plex due to tip geometry and sample inhomogeneity.barrier directly beneath the tip. The upper bound is This model system allowed characterization of realis-determined assuming that a purely resistive current tic tip-sample interactions and lateral sample in-path exists somewhere between the contacts in homogeneity by two dimensional finite elementparallel with the studied potential barrier. This analysis. A relation that was determined numericallyconstant shunting resistor is defined by the resistance was applied to experimental measurements to extractbetween the leads at zero applied bias. For either actual local properties. Extending SSPM to studylimiting case, the local properties of the single grain boundaries of micropatterned polycrystalline

ZnO varistors, the voltage dependence of individualinterfaces has been measured in situ. There is asignificant difference between the macroscopicallyderived barrier response and the voltage dependence

0.5 4 SSPM-local of the individual grain boundary.Ve SSPM-shunt

0.Q4 ' - Macroscopic

0 .Acknowledgements

" ****. •We are grateful to Prof. J. Maier, Max-Planck-0.2 Institute fur Festkorperforschung, Stuttgart, Germany

0 for providing the bicrystal and for helpful discussion.a- 01 This work was supported by the Department of

0 0.5 1 1.5 Energy, Office of Basic Energy Sciences under grantGrain Boundary Applied Bias (Volts) # DEFG02-90ER-45428. The support of the Labora-

tory for Research on the Structure of Matter isFig. 7. SSPM derived local voltage dependence of individual acknowledged for use of the Scanning Probe Facilitygrain boundary potential barrier in Fig. 6 compared to themacroscopic properties that average over hundreds of grainboundaries. The two curves for the local data represent the upper Dominko with microfabrication and of S. Kalininand lower limits due to assumptions regarding the current path. with mathematical analysis is appreciated.

60 B.D. Huey, D.A. Bonnell / Solid State lonics 131 (2000) 51-60

References [14] Frank & Schulte Gmbh, Essen, Germany.[15] M. Leonhardt, Personal Communication, 1998.

[1] T.K. Gupta, Microstructural engineering through donor and [16] 0. Kicnzle, F. Ernst, Effect of shear stress on the atomisticstructure of a grain boundary in strontium titanate, J. Am.acceptor doping in the grain and grain-boundary of a Crm o.8 7 19)13-64Ceram. Soc. 80 (7) (1997) 1639-1644.

polycrystalline semiconducting ceramic, J. Mater. Res. 7 [17] Harris Semiconductor, type V24RA22.(12) (1992) 3280. [18] J. Janata, 100th anniversary of the Kelvin probe. AVSNews-

[2] E. Olsson, G.L. Dunlop, R. Osterlund, Development of letter, 1997(MaylJune): p. 4-5.interfacial microstructure during cooling of a ZnO varistor [19] M. Nonnenmacher, M.P. O'Boyle, H.K. Wickramasinghe,material, J. AppI. Phys. 66 (v. 15) (1989) 5072-5077. Kelvin probe force microscopy, Appl. Phys. Lett. 58 (25)

[3] H.F. Wang, Y.M. Chiang, Thermodynamic stability of inter- (1991) 2921-2923.granular amorphous films in bismuth-doped zinc oxide, J. [20] M. Nonnenmacher, M. Oboyle, H.K. Wickramasinghe, Sur-Am. Ceram. Soc. 81 (1) (1998) 89. face investigations with a Kelvin probe force microscope,

[4] D. Bonnell, I. Solomon, Measurement of space charge Ultramicroscopy 42 (1992) 268.adjacent to oxide grain boundaries by tunneling spectros- [21] H.O. Jacobs, H.F. Knapp, S. Muller, A. Stemmer, Surfacecopy, Ultramicroscopy 42-44 (1992) 788-792. potential mapping: a qualitative material contrast in SPM,

[5] D.L. Carroll, M. Wagner, M. Ruhle, D.A. Bonnell, Schottky Ultramicroscopy 69 (1997) 39-49.barrier formation at nanoscale metal-oxide interfaces, Phys. [22] J.M.R. Weaver, D.W. Abraham, High resolution AFM poten-Rev. B 55 (15) (1997) 9792-9799. tiometry, J. Vac. Sci. Technol. B 9 (3) (1991) 1559-1561.

[6] D.A. Bonnell, B.D. Huey, D.L. Carroll, In situ measurement [23] A.K. Henning, T. Hochwitz, Scanning probe microscopy forof electric fields at individual grain boundaries in TiO_, Solid 2-D semiconductor dopant profiling and device failureState Ionics (1995). analysis, Mater. Sci. Eng. B - Solid State M 42 (1-3)

[7] B.D. Huey, D. Lisjak, D.A. Bonnell, Nanometer scale (1-3)vaitosin interface potential by scnigprobe micro- (96 8variations iscanning [24] EBSP patterns collected in JEOL 6400 SEM using a Merlinscopy, J. Am. Ceram. Soc. (1999) accepted. LTC216-F40E low light camera, a custom Camera Designs

[8] W. Nabhan, B. Equer, A. Broniatowski, G. DeRosny, A Ltd. CCU-1852C camera control unit, and a Hamamatsuhigh-resolution scanning Kelvin probe microscope for con- Argus 10 image processor.tact potential measurements on the 100 nm scale, Rev. Sci. [25] QuickField v. 4.1, Tera Analysis.Instr. 68 (8) (1997) 3108. [26] H.O. Jacobs, P. Leuchtmann, O.J. Homan, A. Stemmer,

[9] T. Trenkler, P. De Wolf, W. Vandervorst, Nanopotentiometry: Resolution and contrast in Kelvin probe force microscopy, J.local potential measurements in complementary metal-oxide- AppI. Phys. D 84 (3) (1998) 1168-1173.semiconductor transistors using atomic force microscopy, J [27] G.D. Skidmore, E.D. Dahlberg, Improved spatial resolutionVac. Sci. Technol. V 16(1) (1998) 367-371. in magnetic force microscopy, Appl. Phys. Lett. 71 (22)

[10] I. Denk, J. Claus, J. Maier, Electrochemical investigations of (1997) 3293-3295.SrTi0 3 boundaries, J. Electrochem. Soc. 144 (10) (1997) [28] D.A. Bonnell, Atomic structure of transition metal oxide3526-3536. surfaces from scanning tunneling microscopy, Progr. Surf.

[11] I. Denk, F. Noll, J. Maier, In situ profiles of oxygen diffusion Sci. 57 (1998) 187-252.in SrTiO 3: Bulk behavior and boundary effects, J. Am. [29] B.D. Huey, D.A. Bonnell, to be published.Ceram. Soc. 80 (2) (1997) 279-285. [30] J.W.P. Hsu, E.B. McDaniel, R.A. Rao, C.B. Eom, Mater. Res.

[12] V. Ravikumar, R.P. Rodrigues, V.P. Dravid, An investigation Soc. 474 (1997) 91-98.of acceptor-doped grain boundaries in SrTiO 3, J. Phys. D:Appl. Phys. 29 (1996) 1799-1806.

[13] K.D. Johnson, V. Dravid, Grain boundary barrier breakdownin niobium donor-doped strontium titanate using in situelectron holography, Appl. Phys. Lett. 74 (4) (1999) 621-623.